Batch Scheduling Research Articles

Qualitatively Analyzing Optimization Objectives in the Design of HPC Resource Manager

A correct evaluation of scheduling algorithms and a good understanding of their optimization criteria are key components of resource management in HPC. In this work, we discuss bias and limitations of the most frequent optimization metrics from the literature. We provide elements on how to evaluate performance when studying HPC batch scheduling. We experimentally demonstrate these limitations by focusing on two use-cases: a study on the impact of runtime estimates on scheduling performance, and the reproduction of a recent high-impact work that designed an HPC batch scheduler based on a network trained with reinforcement learning. We demonstrate that focusing on quantitative optimization criterion (“our work improves the literature by X%”) may hide extremely important caveat, to the point that the results obtained are opposed to the actual goals of the authors. Key findings show that mean bounded slowdown and mean response time are hazardous for a purely quantitative analysis in the context of HPC. Despite some limitations, utilization appears to be a good objective. We propose to complement it with the standard deviation of the throughput in some pathological cases. Finally, we argue for a larger use of area-weighted response time, that we find to be a very relevant objective.

Integrated batch production planning and scheduling with machine-learning based production capacity region considering wind energy system

Continued greenhouse gas emission has led to increased global warming. Towards the goal of carbon emissions reduction and environmental sustainability, replacing high-polluting fossil fuel by clean energy has become a top priority in the decision-making of the current chemical industry. This work presents an optimization framework to address the integration of batch chemical production planning and scheduling considering energy consumption, which is a vital operational problem in chemical production. Assisting the linkage between planning layer and scheduling layer, a novel linear support vector machine based production capacity region calculation method is proposed to identify feasible regions of the scheduling layer. Uncertain factors in both the production system and the energy system are considered, including demand volatility of chemical products, energy supply fluctuations and time-varying energy market prices. To evaluate each strategy and the proposed model, simulation experiments are performed in three representative case studies. The results reveal that the integrated model considering energy consumption shows superior performance in different energy configurations, with or without wind energy generation and battery storage systems.

A clustering-aided multi-agent deep reinforcement learning for multi-objective parallel batch processing machines scheduling in semiconductor manufacturing

Batch processing machines are often the bottleneck in semiconductor manufacturing and their scheduling plays a key role in production management. Pioneer researches on multi-objective batch machines scheduling mainly focus on evolutionary algorithms, failing to meet the online scheduling demand. To deal with the challenges confronted by incompatible job families, dynamic job arrivals, capacitated machines and multiple objectives, we propose a clustering-aided multi-agent deep reinforcement learning approach (CA-MADRL) for the scheduling problem. Specifically, to achieve diverse nondominated solutions, an offline multi-objective scheduling algorithm named Multi-Subpopulation fast elitist Non-Dominated Sorting Genetic Algorithm (MS-NSGA-II) is firstly developed to obtain the Pareto Fronts, and a clustering algorithm based on cosine distance is employed to analyze the distribution of Pareto frontier solution, which would be used to guide reward functions design in multi-agent deep reinforcement learning. To realize multi-objective optimization, several reinforcement learning base models are trained for different optimization directions, each of which composed of batch forming agent and batch scheduling agent. To alleviate time complexity of model training, a parameter sharing strategy is introduced between different reinforcement learning base model. By validating the proposed approach with 16 instances designed based on actual production data from a semiconductor manufacturing company, it has been demonstrated that the approach not only meets the high-frequency scheduling requirements of manufacturing systems for parallel batch processing machines but also effectively reduces the total job tardiness and machine energy consumption.

Automated Optimization of the Synthesis of Alkyl Arenesulfonates in an Undivided Electrochemical Flow Cell

AbstractThe necessary separation of anodic and cathodic compartments in the electrochemical multicomponent synthesis of alkyl arenesulfonates in batch was overcome by the transfer of this reaction in an undivided electrochemical flow cell. The yield was increased from an initial 23 % to 67 % by optimization using Design of Experiments (DoE). The experiments were carried out using an automated experimental flow electrolysis setup controlled by the automation software LABS (Laboratory Automation and Batch Scheduling), an open‐source software that allows to plan and conduct experiments with an arbitrary, freely selectable experimental setup. The automated experimental setup turned out to be stable and provides reproducible results. In total, 6 examples are demonstrated with isolated yields up to 81 %. In addition, the robust scalability of the electrochemical reaction was demonstrated in a 10‐fold scale‐up.

Submodular batch scheduling on parallel machines

AbstractThis article studies a submodular batch scheduling problem motivated by the vacuum heat treatment. The batch processing time is formulated by a monotone nondecreasing submodular function characterized by decreasing marginal gain property. The objective is to minimize the makespan. We show the NP‐hardness of the problem on a single machine and of finding a polynomial‐time approximation algorithm with the worst‐case performance ratio strictly less than for the problem on parallel machines. We introduce a bounded interval to model the batch processing time using two parameters, that is, the total curvature and the quantization indicator. Based on the decreasing marginal gain property and the two parameters, we make a systematic analysis of the full batch longest processing time algorithm and the longest processing time greedy algorithm, and propose the instances with the bound of batch capacity for these two algorithms for the submodular batch scheduling problem. Moreover, we prove the submodularity of batch processing time function of the existing batch models including the parallel batch, serial batch, and mixed batch models. We compare the worst‐case performance ratios in the existing batch models with those deduced from our work in the submodular batch model. In most situations, the worst‐case performance ratios deduced from our work are comparable to the best‐known worst‐case performance ratios with tailored examinations.

Bounded mixed batch scheduling with job release dates and rejection

This paper investigates a bounded mixed batch scheduling problem with job release dates and rejection. The machine processes a batch containing several jobs that their number does not exceed the machine capacity. For the case with a certain release date, we present a polynomial-time exact algorithm. For the case with a constant number of release dates, a pseudo-polynomial-time exact algorithm is proposed. For the general problem, we provide a 2-approximation algorithm and a polynomial-time approximation scheme.

Revitalizing the single batch environment: a ‘Quest’ to achieve fairness and efficiency

In the realm of computer systems, efficient utilization of the CPU (Central Processing Unit) has always been a paramount concern. Researchers and engineers have long sought ways to optimize process execution on the CPU, leading to the emergence of CPU scheduling as a field of study. In this research, we have analyzed the single offline batch processing and investigated other sophisticated paradigms such as time-sharing operating systems and wildly used algorithms, and their shortcomings. Our work is directed toward two fundamental aspects of scheduling: efficiency and fairness. We propose a novel algorithm for batch processing that operates on a preemptive model, dynamically assigning priorities based on a robust ratio, employing a dynamic time slice, and utilizing periodic sorting to achieve fairness. By engineering this responsive and fair model, the proposed algorithm strikes a delicate balance between efficiency and fairness, providing an optimized solution for batch scheduling while ensuring system responsiveness.

Integrated optimization of worker assignment, batch splitting and scheduling for a hybrid assembly line-seru production system

This paper studies an integrated optimization problem of worker assignment, batch splitting and scheduling for a hybrid assembly line-seru production system, which includes multiple serus and a short assembly line. The worker assignment involves assigning each multi-skilled worker to a seru or the assembly line, and assigning the worker tasks to perform. The batch splitting determines the number of subbatches split for each batch, the size of each subbatch, and in which seru the production of each subbatch is performed. The batch scheduling aims to optimize the process sequences of batches in serus and on the assembly line. For the integrated optimization problem, we propose a mixed-integer nonlinear programming model that minimizes the makespan. To solve large-scale instances of the problem, we develop a tailored hybrid variable neighborhood search algorithm with a four-layer solution coding scheme and four novel neighborhood structures. We validate the effectiveness of our model and solution algorithm on two sets of instances. The experimental results indicate that the proposed hybrid metaheuristic achieves a significant reduction in makespan, with an improvement of up to 20.94% compared to the solutions obtained by the Gurobi optimization solver within one hour. In addition, we demonstrate the advantage of optimized batch splitting by comparing it with no batch splitting and equal-size batch splitting strategies in the literature.

A flow shop batch scheduling model with pre-processing and time-changing effects to minimize total actual flow time

A flow shop batch scheduling model with pre-processing and time-changing effects to minimize total actual flow time

A real‐life study on the value of integrated optimization in order picking operations under dynamic order arrivals

AbstractOptimizing the order picking operations is indispensable for warehouses that promise a high customer service level. While many areas for improvement have been identified and studied in the literature, a large gap remains between academia and practice. To help with closing this gap, we perform a case‐study in collaboration with a spare‐parts warehouse in Belgium. In this study, we optimize the order picking operations of the company, using the actual warehouse layout and real order data. A state‐of‐the‐art online integrated order batching, picker routing and batch scheduling algorithm is adapted to consider multiple real‐life constraints. More specifically, the dynamic arrival of new orders is considered, and a capacity constraint on the sorting installation should be respected. Furthermore, a new waiting strategy is studied in which order pickers can temporarily postpone certain orders, as combining them with possible future order arrivals may allow for more efficient overall picking performance. Finally, the performance of the current operating policy is compared with that of both a seed batching heuristic and our metaheuristic algorithm by use of an ANOVA analysis. The results indicate that the number of order pickers can be reduced by 12.5% if the new optimization algorithm is used, accompanied by an improvement in the offered customer service level.

Mixed batch scheduling with non-identical job sizes to minimize makespan

Mixed batch scheduling with non-identical job sizes to minimize makespan

Optimization of heat management and utilization in batch processes

Batch plants boast operational flexibility in manufacturing process and resource sharing, making batch processes suitable to fluctuating-demand or small-scale production. However, given time dependence of operational behavior, batch processes are more demanding in energy integration than continuous processes, a task which is especially important amid the global net zero-emission current. It is, therefore, imperative to develop sustainable design strategies for intermittent processes. The study introduced a targeting method and developed an innovative over-time heat integration technique to optimize energy utilization in batch processes, involving use of heat transfer media for heat recovery over various time intervals and a heat makeup strategy to heat the media over different periods, in order to enhance heat-energy utilization efficiency. The technique is applicable to both fixed and unspecified batch schedules, the latter of which involves simultaneous design of production schedule and heat recovery systems, increasing flexibility and improving heat energy utilization. Case studies are conducted to demonstrate the proposed method. With the heat makeup technique, energy efficiency can be significantly enhanced, achieving energy savings of 51% and 93.1% in hot and cold utility, respectively. Meanwhile, the simultaneous design of production schedule and heat integration can boost savings up to 53% and 99%. Results have proven the benefits of the technique, including significant reduction of energy consumption and carbon emission, contributing to clean production, environmental protection, and sustainable development.

Population-based iterated local search for batch scheduling on parallel machines with incompatible job families, release dates, and tardiness penalties

Population-based iterated local search for batch scheduling on parallel machines with incompatible job families, release dates, and tardiness penalties

A novel segmented random search based batch scheduling algorithm in fog computing

In fog computing, batch scheduling is an important and challenging task aiming at reducing the response latency. Response time includes the scheduling time, execution time and other factors such as network latency. However, the majority of the existing batch scheduling algorithms primarily focus on minimizing the execution time of tasks for IoT devices, ignoring the algorithms' scheduling time. This contributes to suboptimal overall response time, which is a critical quality-of-service (QoS) indicator and significantly impacts performance. Optimizing both execution and scheduling time is therefore crucial for achieving minimum response time, improving QoS, and alleviating load-balancing issues. This work introduces a novel approach for the load-balancing difference variable and uses dimensionality reduction techniques for dynamic task organization. The main contribution comprises two distinctive algorithms: the Pre-allocation Minimum Completion Time (PMT) algorithm and the Segmented Random Search Fog Computing Batch Scheduling Algorithm (SRS). These algorithms are designed to address the inherent characteristics of the problem. To evaluate the performance and applicability of batch scheduling algorithms, we establish a model incorporating compliance and application performance indicators. We introduce the Logical Recursive Indirect Comparison Analysis method to assess and evaluate batch scheduling algorithms. Simulation results demonstrate the exceptional effectiveness of the SRS algorithm. It exhibits a remarkable improvement in comprehensive performance indices, ranging from 139.10% to 261.18%, and optimization quality enhancement rates of 64.25%–154.13% compared to classical standard optimization algorithms. Compared with advanced algorithms, the SRS algorithm also outperforms, with optimization quality improvement rates of 22.74% and 71.11% compared to AEOSSA and CHMPAD.

Optimum Batch Scheduling Model for Quality Aware Delay Sensitive Data Transmission over Fog Enabled IOT Network

The emerging fog networks in the internet of things (IoT) applications provide flexibility and agility for service providers. The combination of fog nodes and edge nodes enable them to deliver a given network service. However, the selection of suitable edge and fog nodes and their scheduling still remain a research challenge. Finding a globally optimal scheduling of oversized data transmission over IoT applications for industrial requirements is crucial. Optimal batch scheduling has been regarded as a viable way to achieve optimal scheduling in other contemporary network models. This manuscript has projected an Optimum Batch Scheduling Model (OBSM) for Quality aware Delay Sensitive Data Transmission over Fog Enabled IoT Networks. A novel clustering technique has been proposed in this manuscript to group the transmission nodes (fog or edge nodes) and data packets, which further pairs each group of data with one of the corresponding node group to achieve delay sensitivity and other quality factors such as energy efficiency. The data scheduling between data and node group is drawn from the previous contribution -"Quality aware Energy Efficient Scheduling Model (QEESM) for Fog Enabled IoT Network". The simulation results have shown that, in terms of average make span rate, average round trip time, and energy consumption, the batch scheduling model OBSM performs noticeably better than the contemporary scheduling models. The OBSM scheduling model's average make-span rate, roundtrip time, as well as energy consumption per make span are 23.3 7.03, 17.8 5.2, and 11.33 6.9 joules, respectively, which conclusively demonstrate that the OBSM model outperforms the existing models. A novel batch scheduling algorithm has been proposed using a unique unsupervised learning approach that suggested to cluster the transmission requests and transmission channels in to multiple clusters.

A two-stage real world serial batching scheduling problem: a case study

This paper concerns a real-life production management problem composed of two stages, which includes interdependent batch scheduling problems. Moreover, both stages require setup operations at the beginning and between the subsequent batches. From the optimisation point of view, this production management problem is complying with customers’ deadlines and hence minimising total lateness. To achieve this objective, the production planner must identify optimum or near-optimum batch schedules for both stages. Correspondingly, this paper aims to develop a methodology to manage the related production problem as accurately as possible. We formulate an optimisation model that employs the mathematical programming method in line with this. Afterwards, an algorithmic proposal based on the simulated annealing algorithm is also developed to solve the problem in realistic sizes. The computational capabilities of the developed model and the algorithm are evaluated on the randomly generated problem sets. A direct comparison between the mathematical model and the algorithm shows how efficiently the proposed algorithm solves real-world problems. Also, computational results indicate that the proposed algorithm satisfactorily solves the related real-world batch scheduling problem.

Improving batch schedulers with node stealing for failed jobs

SummaryAfter a machine failure, batch schedulers typically re‐schedule the job that failed with a high priority. This is fair for the failed job but still requires that job to re‐enter the submission queue and to wait for enough resources to become available. The waiting time can be very long when the job is large and the platform highly loaded, as is the case with typical HPC platforms. We propose another strategy: when a job fails, if no platform node is available, we steal one node from another job , and use it to continue the execution of despite the failure. In this work, we give a detailed assessment of this node stealing strategy using traces from the Mira supercomputer at Argonne National Laboratory. The main conclusion is that node stealing improves the utilization of the platform and dramatically reduces the flow of large jobs, at the price of slightly increasing the flow of small jobs.

Approximation algorithms for mixed batch scheduling on parallel machines

In this paper, we consider two mixed batch scheduling problems on parallel machines, in which multiple jobs can be processed on the same machine simultaneously as a batch, and the processing time of each batch is equal to the weighted sum of the maximum processing time and the total processing time of jobs in this batch. In the first problem, we assume that each job has an inclusive processing restriction, and all the machines have the same capacity. In the second problem, we assume that machines can have different capacities, and there is no inclusive processing restriction for each job. The objectives of both problems are to minimize their makespan. For the first problem, we propose an approximation algorithm whose performance ratio is ( 2 + α ) with 0 ≤ α ≤ 1 . For the second problem, we derive an approximation algorithm with a performance ratio of ( 1 + ρ − ( 1 − α ) / m − α / ( m K min ) ) , where ρ is the ratio of the maximum machine capacity to the minimum machine capacity and K min is the minimum machine capacity.

An improved deep Q-learning algorithm for a trade-off between energy consumption and productivity in batch scheduling

The single-batch machine, commonly found in industrial manufacturing, can concurrently process a group of jobs in variable-speed batches, leading to fluctuating levels of both energy consumption and processing time. Identifying an optimal balance between total energy consumption and makespan presents a challenge due to the intricate nature of their relationship in real-world scenarios. This study introduces a mixed-integer programming model designed to determine optimal machine operating states for diverse workloads, meeting customer requirements while concurrently achieving energy savings for sustainability. An energy-aware batch scheduling deep Q-learning network (EBSDQN) framework has been created, encompassing sequencing policies, batching rules, and speed adjustment policies. This framework is explicitly crafted to tackle the NP-hard problem, a challenge that commonly perplexes commercial solvers like Gurobi when seeking optimal solutions within a 360-second time frame for small-scale instances. The EBSDQN design is fortified with efficient action decoding policies, refined reward assessments, exploitative neighborhood rules, and a streamlined buffer training process. This approach not only conserves training time but also adapts dynamically in accordance with the principles of the Markov Decision Process (MDP) and deep neural network. The developed algorithm exhibits impressive robustness, reaching convergence after 600 episodes of training. In separate comparisons with a commercial solver, two single-objective algorithms, and two multi-objective algorithms, our algorithm consistently demonstrates superior overall performance across the same instances. In the worst-case analysis, further examination delves into the influences of job features. In summary, this study underscores the high sensitivity of both TEC and Cmax to job processing time distribution, suggesting the prioritization of TEC over Cmax in optimization for sustainable planning and operations. Furthermore, this research has the potential to accelerate the integration of artificial intelligence into the manufacturing sector.

Batch Scheduling for Work Centers with Multiple Manufacturing Machines under the Parallel-Sequence-transfer Mode

This paper, based on the parallel-sequence-transfer mode of batch jobs, aimed to implement research on the batch scheduling for work centers with multiple manufacturing machines in full detail. In order to arrange a precise product planning with batch jobs, obtain the maximum completing time with batch jobs, schedule manufacturing machines in each work center accurately, and get the actual number of manufacturing machines used to processing batch jobs in each work center, a new processing procedure of the processing time model under the parallel-sequence-transfer mode was designed. The new designed processing procedure was tested on a numeric example. The results show that the research in this thesis could not only arrange a precise production planning under the parallel-sequence-transfer mode, but also schedule manufacturing machines for each work center with multiple manufacturing machines accurately.